COLLOQUIUM
Department of Computer Science and Engineering
University of South Carolina

A Semantics-based Approach to Machine Perception
Cory Henson
Kno.e.sis  Ohio Center of Excellence in Knowledge-enabled Computing
Wright State University

Date: March 7, 2013
Time: 1430-1530 (2:30pm-3:30pm)
Place: Swearingen 1A03 (Faculty Lounge)

Abstract There are an estimated 40+ billion sensors connected to
mobile devices; and it has been predicted that within the next five
years, sensor data from such devices will become the dominant type of
information on the Web. For this data to be useful in applications,
ranging from healthcare to environmental monitoring, it needs to be
translated into higher-levels of abstraction (e.g., translated from
observed symptoms to disorders). This act of translating low-level
signals into high-level knowledge is called perception, and while
people have evolved sophisticated mechanisms to efficiently perceive
their environment, machines continue to struggle with the task. In
this talk, I will describe a model of machine perception, Intellego,
derived from cognitive theories of perception. Encoded in the Web
Ontology Language (OWL), this model provides a formal semantics of
perception by defining the information processes involved in
collecting and interpreting heterogeneous sensor data. While the use
of OWL enables advanced integration and interpretation of sensor data,
the computational complexity of reasoning seriously limits its
applicability and use within resource-constrained environments, such
as mobile devices. To overcome this challenge, I will discuss an
efficient implementation using bit-vector encodings and operations,
resulting in order-of-magnitude improvements in both efficiency and
scale. The applicability of this approach will be demonstrated through
a real-world application in healthcare. This application, called
kHealth (knowledge-enabled healthcare), provides a means for
patients with chronic heart disease, and their doctors, to remotely
monitor and manage their condition after release from the
hospital. The kHealth application is expected to go into pre-clinical
trials by the end of March.

Cory Henson is a Ph.D. Candidate at Wright State University and a
researcher at Kno.e.sis, the Ohio Center of Excellence in
Knowledge-enabled computing. He also works at Riverside Research, a
research organization spun out of Columbia University, where he
collaborates on projects for the DoD and Intelligence Community. He
has bachelor's degrees in both Computer Science and Cognitive
Science from the University of Georgia. Cory's research revolves
around Web-based data and metadata management, including knowledge
representation, ontology modeling, data integration, and
reasoning. More specifically, his research comprises the synergistic
use of semantic web, artificial intelligence, and cognitive science to
develop the theoretical foundation and practical tools needed to
enable sensor data management and machine perception. This technology
has exciting applications in developing smart environments and
enabling more proactive, preventative healthcare. For more, see:
http://knoesis.org/researchers/cory.